Yes, that's right. This is not a practical approach given current LLM limitations. But perhaps in the future if the cost of each individual call decreases, and if latency decreases (and models run locally), but context window remains limited this approach could be more useful as it uses pairwise comparisons rather than loading the entire context window.
This essay by Huxley said exactly what I have always felt but never articulated. The problem is far worse today than it was in his time. So much of recreation today consists of binge-watching Netflix. In addition, because Netflix has global reach, the world often watches the same content, devoid of local meaning.
I have realized that active recreation, although requiring overcoming a certain threshold of activation energy, is far more compelling and fulfilling than passive recreation.